how can I say to python pd to aply something to all the columns except the ones with strings?-CodePudding

I have a pd

    A       B       C    D
0   1,598   65.79   79   asdf
1   -300    46.90   90   qwer

how can I say to aply this: `

dff = dff.astype('str').apply(lambda x: pd.to_numeric(x.str.replace(',', '')))

` to only the columns with ","'s?

take into consider that I don't know the keys of the columns or orders

I don't know very well how to approach this

CodePudding user response：

you can check numeric columns or object columns that all values in column can convert to numeric

df.astype('str').apply(lambda x: x.str.contains(r'^[0-9-.,] $')).all()

output:

A     True
B     True
C     True
D    False
dtype: bool

A, B, C

you can filter columns by boolean indexing

cond1 = df.astype('str').apply(lambda x: x.str.contains(r'^[0-9-.,] $')).all()
df.loc[:, cond1]

output:

    A       B       C
0   1,598   65.79   79
1   -300    46.90   90

you can convert only that columns to numeric

cols = df.loc[:, cond1].columns
df[cols] = df[cols].astype('str').apply(lambda x: pd.to_numeric(x.str.replace(',', '')))
df

output:

    A       B       C   D
0   1598    65.79   79  asdf
1   -300    46.90   90  qwer

you can check data type of result:

df.dtypes

A      int64
B    float64
C      int64
D     object
dtype: object

A, B and C columns convert to numeric, D column is still object